Description
When dealing with benchmarking, result collection and sharing is as essential as the production of the data itself. A central source of information allows for data comparison, visualization, and analysis, which the community both contributes to and profits from.
While this is the case in other fields, in the High Energy Physics (HEP) benchmarking community both script and result sharing required human interaction in the past, which greatly hindered data collection and retention.
Thus, in the context of the HEP Benchmark Project, a tool has been devised to unify the benchmark execution process as well as the format of the reported results: the HEP Benchmark Suite. The Suite is complemented by an OpenSearch instance, allowing for additional functionalities such as long-term storage, visualization, and monitoring.
This state-of-the-art solution aims at solving this challenge present in the HEP community by becoming its benchmarking cornerstone for all HEP sites around the globe. Its architectural design will be described in this contribution, together with the experience acquired while gathering thousands of measurements worldwide across the WLCG.
| Key Topic | Data analytics platforms and reproducible open science |
|---|---|
| Other key topic | Benchmarking |
